SHORT REPORT

Emotion words and categories: evidence from lexical decision

Graham G. Scott • Patrick J. O’Donnell •

Sara C. Sereno

Received: 29 April 2013 / Accepted: 5 November 2013

� Marta Olivetti Belardinelli and Springer-Verlag Berlin Heidelberg 2013

Abstract We examined the categorical nature of emo-

tion word recognition. Positive, negative, and neutral

words were presented in lexical decision tasks. Word fre-

quency was additionally manipulated. In Experiment 1,

‘‘positive’’ and ‘‘negative’’ categories of words were

implicitly indicated by the blocked design employed. A

significant emotion–frequency interaction was obtained,

replicating past research. While positive words consistently

elicited faster responses than neutral words, only low fre-

quency negative words demonstrated a similar advantage.

In Experiments 2a and 2b, explicit categories (‘‘positive,’’

‘‘negative,’’ and ‘‘household’’ items) were specified to

participants. Positive words again elicited faster responses

than did neutral words. Responses to negative words,

however, were no different than those to neutral words,

regardless of their frequency. The overall pattern of effects

indicates that positive words are always facilitated, fre-

quency plays a greater role in the recognition of negative

words, and a ‘‘negative’’ category represents a somewhat

disparate set of emotions. These results support the notion

that emotion word processing may be moderated by dis-

tinct systems.

Keywords Emotion � Word frequency � Category �Lexical decision � Arousal � Valence

Introduction

Recent word recognition research has reported an interaction

between a word’s emotional quality (characterized as posi-

tive, negative, or neutral) and its frequency of occurrence

(having a higher or lower prevalence of use). These results

were found in lexical decision reaction times (Kuchinke et al.

2007; Scott et al. 2009), in electrophysiological voltages

(Scott et al. 2009), as well as in eye fixation times during

fluent reading (Scott et al. 2012). Specifically, for low fre-

quency (LF) words, behavioral responses to both positive and

negative words were faster than those to neutral words; for

high frequency (HF) words, responses to positive words

alone were faster than those to either negative or neutral

words (which did not differ). Early word frequency effects

have consistently been demonstrated in eye movement and

electrophysiological paradigms (see Hand et al. 2010), and

are considered to reliably indicate lexical access (e.g., Sereno

and Rayner 2003). Thus, an interaction of a word’s emotional

quality with its frequency suggests a central role of emotion in

the initial stages of word recognition.

The underlying theoretical mechanisms of emotion word

processing, however, are less well understood. One account

is derived from Taylor’s (1991) two-stage mobilization-

minimization hypothesis, developed from McGinnies’

(1949) theory of perceptual defense (see also Pratto and

John’s (1991) automatic vigilance hypothesis). Because of

their high arousal, emotion words are initially facilitated

relative to neutral words. Potential negative consequences of

such emotional content, however, are guarded against by

delaying their processing to provide time to diminish their

impact. Accordingly, although both positive and negative

words enjoy an initial advantage, negative words are sub-

sequently inhibited. Scott et al. (2009) further suggested that

minimization could be stronger for HF than LF negative

G. G. Scott

Division of Psychology, School of Social Sciences, University of

the West of Scotland, Paisley PA1 2BE, UK

P. J. O’Donnell � S. C. Sereno (&)

Institute of Neuroscience and Psychology, School of

Psychology, University of Glasgow, 58 Hillhead Street,

Glasgow G12 8QB, UK

e-mail: Sara.Sereno@glasgow.ac.uk

123

Cogn Process

DOI 10.1007/s10339-013-0589-6

words because, by definition, HF concepts are more salient.

An alternative explanation for the differential pattern of

responses to HF and LF negative words is based on the

clinical notion of desensitization. In their ‘‘boy who cried

wolf’’ hypothesis, Scott et al. (2012) proposed that the rela-

tive slowness of responses to HF negative words may be

because their negative semantics are diluted or lost through

repeated exposure. Thus, while such words are consciously

considered as negative in off-line rating tasks, on-line task

performance may be more closely linked to automatic word

recognition processes in which only vestigial emotional

activations are elicited.

Emotion words are typically characterized by their dual

properties of arousal (internal activation) and valence (value

or worth). In comparison with neutral words, emotion words

have high arousal values correlated with extreme valence

(e.g., Bradley and Lang 1999; see also the circumplex model

of Russell 1980). While emotion words reside at polar

opposites of a valence continuum, the question remains as to

whether positive and negative words comprise a single

‘‘emotion’’ category or form independent categories. For

example, some researchers suggest that the relationship

between valence and recognition is linear, extending over a

single dimension (e.g., Kousta et al. 2009; Larsen et al.

2008), while others maintain it is categorical, with distinct

positive and negative types (e.g., Estes and Adelman 2008a,

b). Research into the organization and representation of

categories has demonstrated selective facilitation of cate-

gory members across a variety of paradigms and measures

(e.g., Bermeitinger et al. 2011; Sachs et al. 2008; Segalowitz

and Zheng 2009). Moreover, what defines a category (e.g.,

Barsalou 1983) and whether a category is established

implicitly or explicitly, for example, via the context afforded

by a list of related items or by an encompassing label (e.g.,

Bazzanella and Bouquet 2011; Becker 1980; Schacter and

Badgaiyan 2001), has implications for the amount of benefit

conferred on its members.

To investigate the categorical nature of emotion word

processing, we designed a series of lexical decision experi-

ments that all additionally manipulated word frequency, in

particular, because of its differential effect on responses to

negative words. In the prior emotion 9 frequency lexical

decision studies, positive and negative words were inter-

mixed with neutral words within a single block. In our first

study (Experiment 1), we examined whether the same pattern

of effects would be obtained under conditions of implicit

categorical priming—when positive and negative words

were presented in separate blocks. In the subsequent two

studies, we examined the effect of explicit category priming,

comparing responses to words belonging to the neutral cat-

egory of ‘‘household’’ items to those within the category of

either ‘‘positive’’ (Experiment 2a) or ‘‘negative’’ (Experi-

ment 2b) items. In the Kuchinke et al. (2007) and Scott et al.

(2009) experiments (as in most emotion word experiments

and in our Experiment 1), neutral words did not form a

coherent category; they were simply items that shared the

characteristics of low arousal and intermediate valence.

Thus, it is possible that the response time advantage found

for (most) emotion over neutral words may be due in part to

unbalanced implicit categorical priming across conditions,

where a greater degree of semantic links exist between a

selection of emotion versus neutral words. Consequently,

while explicit category priming should facilitate the pro-

cessing of all word types, this effect may appear more pro-

nounced for neutral words. We expected that the current set

of experiments would provide complementary results. That

is, Experiment 1 should establish a baseline of implicit cat-

egorical emotion processing (positive and negative) relative

to neutral words that do not form any coherent category.

Experiments 2a and 2b should provide evidence of the effect

of explicit categorical priming when it has been applied to all

conditions (emotional or not). Taken together, the results of

these experiments should begin to address the role that cat-

egorical priming plays in experimental research into emotion

word processing.

Experiment 1

Method

Participants

Twenty-four members of the University of Glasgow com-

munity (16 female; mean age 25) received compensation for

their participation. All were native English speakers, were

right-handed, had normal or corrected-to-normal vision, and

were naıve as to the purpose of the experiment.

Apparatus

The experiment was run on a Mac G4 using PsyScope 1.2.5

PPC software. Stimuli were presented in 24-point Courier

(black characters on a white background) on a Hansol 2100A

1900 monitor. At a viewing distance of 2500, 3 characters

subtended 1o of visual angle. Responses were made via a

PsyScope Button Box, and reaction times (RTs) were

recorded with millisecond accuracy.

Design and materials

A 4 (Emotion: Positive, Negative, Neutral1, Neutral2) 9 2

(Frequency: LF, HF) within-participants design was used

with 27 items in each of the 8 conditions. Arousal and

valence ratings for all words were acquired from the affec-

tive norms for English words (ANEW) database (Bradley

Cogn Process

123

and Lang 1999). Each word has associated ratings for

arousal, from 1 (low) to 9 (high), and for valence, from 1

(low, having a negative meaning) to 9 (high, having a posi-

tive meaning). Arousal values ranged from 6 to 9 for emotion

words and 1 to 5.5 for neutral words. Valence values ranged

from 6 to 9 for positive words, 1 to 4 for negative words, and

4 to 6 for neutral words. Word frequencies were obtained

from the British National Corpus (BNC; http://www.

natcorp.ox.ac.uk), a database of 90 million written word

tokens (with frequencies expressed in occurrences per mil-

lion). Example items are presented in Table 1. Average word

frequency, length (in letters and syllables), arousal, and

valence values across conditions are presented in Table 2.

For each word, a nonword of equal length was constructed.

Nonwords comprised pronounceable, orthographically legal

pseudowords (e.g., famper).

Procedure

Participants were tested individually. Word responses were

made using the right forefinger on the right (green) key of the

Button Box, labeled ‘‘W,’’ and nonword responses with the

left forefinger on the left (red) key, labeled ‘‘NW.’’ Partici-

pants were first presented with a set of practice items to

become accustomed to the task.

The experiment comprised two blocks, with Positive and

Neutral1 words in one block, and Negative and Neutral2

words in the other. Half of the participants received the

Positive and Netural1 block first, while the other half

received the Negative and Neutral2 block first. Within each

block, trials were presented in a different random order for

each participant. Each trial consisted of a blank screen

(1,000 ms), a central fixation cross (200 ms), another blank

screen (500 ms), and a letter string presented centrally (until

response).

Results

The RTs from correct responses (96 % of the data) were

subjected to two trimming procedures. Items with RTs less

than 250 ms or greater than 1,500 ms were excluded from

further analyses. For each participant in each condition,

items with RTs beyond two standard deviations of the mean

were additionally excluded. These procedures resulted in an

average data loss of 5 % per participant (approx. one item per

condition).

The mean RT data are presented in Table 3 and are

graphically depicted in Fig. 1. A 4 (Emotion) 9 2 (Fre-

quency) analysis of variance (ANOVA) was performed on

the data by participants (F1) and by items (F2). There were

significant main effects of Emotion and Frequency as well as

a significant Emotion 9 Frequency interaction [Emotion:

F1(3,69) = 6.83, p \ .001, and F2(3,78) = 12.77,

p \ .001; Frequency: F1(1,23) = 114.89, p \ .001, and

F2(1,26) = 146.84, p \ .001; Emotion 9 Frequency:

F1(3,69) = 6.86, p \ .001, and F2(3,78) = 4.11, p = .009].

Follow-up contrasts to the interaction demonstrated signifi-

cant effects of Frequency, with faster responses to HF than

LF words, for all Emotion conditions [Fs [ 7.09, ps \ .01].

For LF words, RTs to both Positive and Negative words,

which did not differ [Fs \ 1], were significantly faster than

those to either Neutral1 or Neutral2 words [Fs [ 15.35,

ps \ .001], which did not differ [F1 = 1.14, p [ .25;

F2 \ 1]. For HF words, responses to Positive words alone

were significantly faster than responses to Negative, Neu-

tral1, or Neutral2 words, although some of these effects were

marginal by items [F1s [ 7.19, ps \ .01; F2s from 4.00 to

3.66, ps from .049 to .060]. HF Neutral1, Neutral2, and

Negative conditions did not differ [Fs \ 1].

Discussion

HF and LF positive, negative, and neutral words were pre-

sented in a lexical decision task. Unlike the prior emotion–

frequency studies, positive and negative words appeared in

separate blocks. Nonetheless, a similar pattern of results

emerged showing a significant interaction. It could be that

the implicit category structure of ‘‘positive’’ and ‘‘negative’’

(separate blocks) is just as effective as that of ‘‘emotion’’

(single block). A simpler explanation for these findings is

that the emotion–frequency interaction is immune to rela-

tively weak contextual manipulations.

Experiments 2a and 2b

In Experiments 2a and 2b, a stronger semantic context was

implemented by providing explicit category labels. Specifi-

cally, participants were told that they would be presented

Table 1 Example materials from Experiments 1, 2a, and 2b

Frequency

LF HF

Experiment 1

Positive cheer, miracle, treasure cash, travel, victory

Negative snake, outrage, mutilate fire, cancer, hostile

Neutral1 muddy, lantern, highway tower, finger, museum

Neutral2 salad, basket, hairpin clock, writer, square

Experiment 2a

Positive flirt, reunion, valentine cash, passion, birthday

Household stove, hammer, hairdryer door, window, corridor

Experiment 2b

Negative shark, slave, terrorist bomb, panic, disaster

Household stool, poster, appliance bowl, bench, bathroom

LF low frequency, HF high frequency

Cogn Process

123

with words belonging to two categories: ‘‘positive’’ and

‘‘household’’ items (Experiment 2a) or ‘‘negative’’ and

‘‘household’’ items (Experiment 2b).

Method

Participants

Thirty-six members of the University of Glasgow commu-

nity were compensated for their participation—18 (15

female; mean age 19) in Experiment 2a and a different set of

18 (17 female; mean age 20) in Experiment 2b. None had

participated in Experiment 1. All conformed to the same

criteria used in Experiment 1.

Apparatus

The apparatus was identical to that of Experiment 1.

Design and Materials

Both experiments utilized a 2 (Category: emotional, neu-

tral) 9 2 (Frequency: LF, HF) within-participants design,

with 18 items in each of the 4 conditions. The emotional

category was ‘‘Positive’’ items in Experiment 2a and ‘‘Neg-

ative’’ items in Experiment 2b, while the neutral category was

‘‘Household’’ items in both experiments. Emotion and neutral

words were selected from ANEW within the same ranges of

arousal and valence values used in Experiment 1. Example

items are presented in Table 1 and average stimulus proper-

ties in Table 2. For each experiment, nonwords employing

the same criteria as in Experiment 1 were used.

Procedure

The procedure of Experiment 1 was used with two modifi-

cations. First, as part of their instructions, participants were

informed that the words consisted of items from two cate-

gories—‘‘Positive’’ and ‘‘Household’’ items (Experiment

2a) or ‘‘Negative’’ and ‘‘Household’’ items (Experiment 2b).

Second, items were presented within a single block.

Results

The RTs from correct responses (96 % in both experi-

ments) were subjected to the trimming procedures of

Table 2 Specifications of materials in Experiments 1, 2a, and 2b

Frequency Letters Syllables Arousal Valence

LF HF LF HF LF HF LF HF LF HF

Experiment 1

Positive 7.1 56.9 6.9 6.0 2.2 1.9 6.6 6.6 7.7 7.8

Negative 6.7 56.4 6.7 5.8 2.0 1.8 6.6 6.8 2.6 2.6

Neutral1 6.2 57.9 6.6 5.8 2.0 1.9 4.3 3.7 5.1 5.4

Neutral2 6.4 54.9 6.6 5.9 2.1 1.8 3.8 4.1 5.3 5.1

Experiment 2a

Positive 5.9 56.4 6.4 5.8 2.0 1.7 6.7 6.6 7.7 7.8

Household 5.8 60.0 6.4 5.7 2.1 1.7 3.8 3.8 5.2 5.2

Experiment 2b

Negative 7.5 56.6 6.6 5.6 1.9 1.9 6.7 6.9 2.6 2.4

Household 6.0 63.7 6.7 5.6 2.1 1.7 3.9 3.7 5.2 5.3

Units of measurement are as follows: frequency in occurrences per million, word length in number of letters and syllables, arousal on a scale

from 1 (low) to 9 (high), valence on a scale from 1 (low, having a negative meaning) to 9 (high, having a positive meaning)

LF low frequency, HF high frequency

Table 3 Mean reaction time (RT) in milliseconds (with standard

deviations) across conditions in Experiments 1, 2a, and 2b

Frequency

LF HF

Experiment 1

Positive 525 (81) 483 (63)

Negative 526 (81) 502 (82)

Neutral1 558 (96) 501 (72)

Neutral2 565 (81) 502 (65)

Experiment 2a

Positive 576 (97) 533 (84)

Household 597 (105) 559 (119)

Experiment 2b

Negative 581 (103) 517 (74)

Household 575 (80) 516 (75)

LF low frequency, HF high frequency

Cogn Process

123

Experiment 1, resulting in an average data loss of 5 % per

participant (approx. one item per condition). RT means are

presented in Table 3 and Fig. 1. For each experiment, 2

(Category) 9 2 (Frequency) ANOVAs (F1 and F2) were

performed on the RT means.

In Experiment 2a, the main effects of Category and Fre-

quency were both significant [Category: F1(1,17) = 7.82,

p = .012, and F2(1,17) = 7.94, p = .012; Frequency:

F1(1,17) = 16.94, p \ .001, and F2(1,17) = 12.31,

p = .003]. Responses to ‘‘positive’’ items (555 ms) were

faster than those to ‘‘household’’ items (578 ms). In addi-

tion, faster responses were made to HF (546 ms) than to

LF (586 ms) words. There was no evidence of an interaction

[all Fs \ 1].

In Experiment 2b, only the main effect of Frequency

was significant [Frequency: F1(1,17) = 40.59, p \ .001,

and F2(1,17) = 36.29, p \ .001]. Responses to HF words

(516 ms) were faster than those to LF words (578 ms).

There was no effect of category nor was the interaction

significant [all Fs \ 1].

Discussion

Experiments 2a and 2b explicitly manipulated category

membership for both emotion and neutral items. In both

experiments, significant word frequency effects were

obtained, with faster responses to HF than LF words.

‘‘Positive’’ words elicited faster responses than ‘‘House-

hold’’ words (Experiment 2a), whereas ‘‘Negative’’ words

were no different (Experiment 2b). In comparison with the

pattern of results of Experiment 1, the relative relationship

between positive and neutral words remained the same,

while that between negative and neutral words changed.

Specifically, the previous advantage within the LF condi-

tion of negative over neutral words disappeared. Accord-

ingly, it seems that category priming similarly affected

positive and neutral, but not negative, words. One expla-

nation draws on the differential effects of category priming

in relation to both word frequency and category breadth,

discussed in greater depth in the next section.

General discussion

We sought to determine the role of implicit and explicit

category membership on lexical decision responses to

positive, negative, and neutral words. Word frequency was

additionally manipulated because of its central relationship

with lexical access. Prior investigations demonstrated an

emotion–frequency interaction using different paradigms

and measures (Kuchinke et al. 2007; Scott et al. 2009,

2012). These studies share the methodological feature that

all word types were presented together. It is possible that

emotion words (positive and negative) were selectively

facilitated in their processing because implicit categorical

priming was present for emotion but not neutral words.

In Experiment 1, positive and negative words were

embedded in separate blocks, with neutral words included

in each block. We reasoned that implementing such a

design should help activate the implicit categories of

‘‘positive’’ and ‘‘negative’’ rather than that of ‘‘emotion.’’

Nevertheless, as with the single-block studies, a similar

emotion–frequency interaction was obtained. All word

types demonstrated frequency effects. For LF words,

positive and negative word responses were faster than

neutral word responses; for HF words, only positive word

response were faster than negative and neutral responses. It

is possible that the benefit provided by two focal implicit

categories offset the cost of eliminating a single general

one. Indeed, a recent study that only employed a subset of

our conditions—HF and LF negative and neutral words—

reported an identical pattern of results corresponding to

these conditions (Mendez-Bertolo et al. 2011; cf. Nakic

Fig. 1 Average RT (ms) in Emotion 9 Frequency conditions in Experiments 1, 2a, and 2b. LF low frequency, HF high frequency

Cogn Process

123

et al. 2006). It is also possible that although positive and

negative items were blocked in the current study, partici-

pants nonetheless relied upon a more general implicit

category. More parsimoniously, the effect of such implicit

category priming could be too weak to sufficiently influ-

ence word recognition processes. This is substantiated by

the Scott et al. (2012) study which demonstrated a similar

emotion–frequency interaction in eye fixation times on

target words during fluent reading. Because targets were

embedded in emotionally neutral sentence frames, the word

recognition process was less susceptible to local priming

by semantically related exemplars.

Experiments 2a and 2b examined whether explicit cat-

egory priming of both emotion and neutral words would

alter their response–time relationships. Participants were

instructed that word stimuli belonged to two categories—

‘‘positive’’ and ‘‘household’’ items (Experiment 2a), or

‘‘negative’’ and ‘‘household’’ items (Experiment 2b). Word

frequency effects were obtained across all word categories.

‘‘Positive’’ words elicited consistently faster responses than

‘‘household’’ words, while no differences emerged for

‘‘negative’’ versus ‘‘household’’ words. What distinguishes

these combined results from those of the prior emotion–

frequency experiments is the lack of a difference between

LF negative and neutral words. It has been demonstrated

that semantic variables, including category priming, often

exert a greater facilitative effect on LF than HF words (e.g.,

Becker 1979; Hauk et al. 2006; for reviews, see Borowsky

and Besner 2006; Hand et al. 2010). As one consequence,

the magnitude of word frequency effects should be

reduced. Our current study, however, does not address this

particular aspect of the data as independent samples of

participants were used. Nevertheless, it is possible that

explicit category priming, with particular benefits to LF

words, is selectively effective for positive and neutral, but

not negative, categories. While all three categories are

fairly broad, the negative category is perhaps the most

heterogeneous and, hence, the least susceptible to such

priming. In their density hypothesis, Unkelbach et al.

(2008) proposed that positive information is processed

more quickly because, in comparison with negative infor-

mation, it is more densely clustered in semantic space.

Emotions have typically been classified into the subtypes

of ‘‘happiness,’’ ‘‘surprise,’’ ‘‘sadness,’’ ‘‘anger,’’ ‘‘fear,’’

and ‘‘disgust’’ (e.g., Ekman and Friesen 1971). In com-

parison with the category ‘‘positive,’’ the category ‘‘nega-

tive’’ comprises a greater range of distinct emotions. For

example, while many negative words will engage the

activation of an avoidance mechanism, words that belong

to the category ‘‘anger’’ often involve approach actions.

Moreover, depending on the perspective (subject/object)

taken by the reader, a word’s interpretation can be influ-

enced by their appraisal (Lerner and Keltner 2000).

Theoretically, to more effectively assess explicit cate-

gorical priming of emotional words, it might be beneficial to

include additional baseline conditions (e.g., HF and LF non-

categorical neutral words) as well as discrete subcategories

of emotion words (e.g., HF and LF ‘‘happy’’ and ‘‘anger’’

words). On the practical side, however, implementing such

control conditions within a single experiment would be

methodologically challenging and the results could be

problematic to interpret. First, it would be difficult to gen-

erate an adequate number of items across all conditions that

are controlled for relevant lexical properties. Second,

responses to a minority of non-categorical items within the

broader context of a task involving categories may not be

representative. Finally, it is unclear whether words belong-

ing to categories of distinct emotions (e.g., ‘‘happy’’ and

‘‘anger’’ words) would be additionally considered as mem-

bers of the generic positive and negative word categories.

Nevertheless, there may be other ways of addressing con-

cerns about what can and cannot be directly attributable to

explicit category priming of emotional words.

Recently, a growing minority of researchers have begun

to investigate distinct emotional subcategories underpinning

the meaning of words (e.g., Briesemeister et al. 2011a, b;

Fontaine et al. 2007; Stevenson et al. 2007; Wurm 2007). For

example, Briesemeister et al. (2011a) showed that a word’s

membership within certain discrete emotional categories

(‘‘happiness,’’ ‘‘disgust,’’ and ‘‘fear,’’ but not ‘‘anger’’ or

‘‘sadness’’) could explain as much variance in lexical deci-

sion RTs as a two-dimensional (arousal 9 valence) model of

emotion. Although word frequency was statistically con-

trolled via regression, interactions with frequency were not

tested. One experiment was in German and used a set of

affectively laden words. However, the English data from the

other experiments were taken from the English Lexicon

Project (Balota et al. 2007), representing responses to over 40

thousand words collected from many experiments across

several labs (different stimulus lists comprised nonadjacent

words from an alphabetized master list). Consequently, it is

difficult to ascertain the semantic features of any subset of

materials, in particular, whether any form of categorical

priming could have occurred. Further studies examining the

categorical nature of emotion word processing should use

appropriate neutral words having comparable categorical

characteristics.

In sum, we examined the categorical nature of emotion

word processing in a series of experiments. When implicit

categories of ‘‘positive’’ and ‘‘negative’’ were implied

(Experiment 1), the emotion–frequency interaction found in

prior studies was maintained. When explicit categories were

employed (Experiments 2a and 2b), including one repre-

sentative of neutral words, only positive words retained their

relative behavioral advantage over neutral words; responses

to LF negative words lost their prior advantage and were no

Cogn Process

123

different than their neutral counterparts. We suggested that

the pattern of results might be explained by effects of cate-

gorical priming that depend on both word frequency and

category coherence. In this respect, our work represents an

initial exploration of these issues. Recent developments in

establishing larger and more comprehensive databases of

words that are normed on several emotional dimensions are

encouraging. Such work will permit more systematic

assessments of how a word’s emotional content as well as its

frequency and context can affect its recognition.

References

Balota DA, Yap MJ, Cortese MJ, Hutchison KA, Kessler B, Loftis B,

Neely JH, Nelson DL, Simpson GB, Treiman R (2007) The

English Lexicon Project. Behav Res Methods 39(3):445–459

Barsalou LW (1983) Ad hoc categories. Mem Cogn 11(3):211–227

Bazzanella B, Bouquet P (2011) Associative and categorical priming

in recognition of individuals. In: Carlson L, Holscher C, Shipley

TF (eds) Proceedings of the 33rd Annual Conference of the

Cognitive Science Society. Cognitive Science Society, Boston,

MA, pp 525–530

Becker CA (1979) Semantic context and word frequency effects in

visual word recognition. J Exp Psychol Hum Percept Perform

5(2):252–259

Becker CA (1980) Semantic context effects in visual word recogni-

tion: an analysis of semantic strategies. Mem Cogn 8(6):493–512

Bermeitinger C, Wentura D, Frings C (2011) How to switch on and

switch off semantic priming effects for natural and artifactual

categories: activation processes in category memory depend on

focusing specific feature dimensions. Psychon Bull Rev

18(3):579–585

Borowsky R, Besner D (2006) Parallel distributed processing and

lexical-semantic effects in visual word recognition: are a few

stages necessary? Psychol Rev 113(1):181–195

Bradley MM, Lang PJ (1999) Affective Norms for English Words

(ANEW): Stimuli, instruction manual, and affective ratings.

Technical Report C-1. The Center for Research in Psychophys-

iology, University of Florida, Gainsville, FL

Briesemeister BB, Kuchinke L, Jacobs AM (2011a) Discrete emotion

effects on lexical decision response times. PLoS ONE

6(8):e23743

Briesemeister BB, Kuchinke L, Jacobs AM (2011b) Discrete Emotion

Norms for Nouns: Berlin Affective Word List (DENN-BAWL).

Behav Res Methods 43(2):441–448

Ekman P, Friesen WV (1971) Constants across cultures in the face

and emotion. J Pers Soc Psychol 17(2):124–129

Estes Z, Adelman JS (2008a) Automatic vigilance for negative words

in lexical decision and naming: comment on Larsen, Mercer, and

Balota (2006). Emotion 8(4):441–444

Estes Z, Adelman JS (2008b) Automatic vigilance for negative words

is categorical and general. Emotion 8(4):453–457

Fontaine JRJ, Scherer KR, Roesch EB, Ellsworth PC (2007) The

world of emotion is not two-dimensional. Psychol Sci

18(12):1050–1057

Hand CJ, Miellet S, O’Donnell PJ, Sereno SC (2010) The frequency-

predictability interaction in reading: it depends where you’re coming

from. J Exp Psychol Hum Percept Perform 36(5):1294–1313

Hauk O, Davis MH, Ford M, Pulvermuller F, Marslen-Wilson WD

(2006) The time course of visual word recognition as revealed by

linear regression analysis of ERP data. NeuroImage

30(4):1383–1400

Kousta S-T, Vinson DP, Vigliocco G (2009) Emotion words,

regardless of polarity, have a processing advantage over neutral

words. Cognition 112(3):473–481

Kuchinke L, Vo ML, Hofmann M, Jacobs AM (2007) Pupillary

responses during lexical decisions vary with word frequency but

not emotional valence. Int J Psychophysiol 65(2):132–140

Larsen RJ, Mercer KA, Balota DA, Strube MJ (2008) Not all negative

words slow down lexical decision and naming speed: importance

of word arousal. Emotion 8(4):445–452

Lerner JS, Keltner D (2000) Beyond valence: toward a model of

emotion-specific influences on judgement and choice. Cogn

Emot 14(4):473–493

McGinnies E (1949) Emotionality and perceptual defense. Psychol

Rev 56(5):244–251

Mendez-Bertolo C, Pozo JA, Hinojosa JA (2011) Word frequency

modulates the processing of emotional words: convergent

behavioral and electrophysiological data. Neurosci Lett

494(3):250–254

Nakic M, Smith BW, Busis S, Vythilingam M, Blair RJR (2006) The

impact of affect and frequency on lexical decision: the role of the

amygdala and inferior frontal cortex. NeuroImage

31(4):1752–1761

Pratto F, John OP (1991) Automatic vigilance: the attention-grabbing

power of negative social information. J Pers Soc Psychol

61(3):380–391

Russell JA (1980) A circumplex model of affect. J Pers Soc Psychol

39(6):1161–1178

Sachs O, Weis S, Zellagui N, Huber W, Zvyaginstev M, Mathiak K,

Kircher T (2008) Automatic processing of semantic relations in

fMRI: neuronal activation during semantic priming of taxonomic

and thematic categories. Brain Res 1218:194–205

Schacter DL, Badgaiyan RD (2001) Neuroimaging of priming: new

perspectives on implicit and explicit memory. Curr Dir Psychol

Sci 10(1):1–4

Scott GG, O’Donnell PJ, Leuthold H, Sereno SC (2009) Early

emotion word processing: evidence from event-related poten-

tials. Biol Psychol 80(1):95–104

Scott GG, O’Donnell PJ, Sereno SC (2012) Emotion words affect eye

fixations during reading. J Exp Psychol Learn Mem Cogn

38(3):783–792

Segalowitz SJ, Zheng X (2009) An ERP study of category priming:

evidence of early lexical semantic access. Biol Psychol

80(1):122–129

Sereno SC, Rayner K (2003) Measuring word recognition in reading:

eye movements and event-related potentials. Trends Cogn Sci

7(11):489–493

Stevenson RA, Mikels JA, James TW (2007) Characterization of the

Affective Norms for English Words by discrete emotional

categories. Behav Res Methods 39(4):1020–1024

Taylor SE (1991) Asymmetrical effects of positive and negative

events: the mobilization-minimization hypothesis. Psychol Bull

110(1):67–85

Unkelbach C, Fiedler K, Bayer M, Stegmuller M, Danner D (2008)

Why positive information is processed faster: the density

hypothesis. J Pers Soc Psychol 95(1):36–49

Wurm LH (2007) Danger and usefulness: an alternative framework

for understanding rapid evaluation effects in perception?

Psychon Bull Rev 14(6):1218–1225

Cogn Process

123